ARTICLE

Role of habitat complexity in predator–prey dynamics betweenan introduced fish and larval Long-toed Salamanders(Ambystoma macrodactylum)Erin K. Kenison, Andrea R. Litt, David S. Pilliod, and Tom E. McMahon

Abstract: Predation by nonnative fishes has reduced abundance and increased extinction risk for amphibian populationsworldwide. Although rare, fish and palatable amphibians have been observed to coexist where aquatic vegetation and structuralcomplexity provide suitable refugia. We examined whether larval Long-toed Salamanders (Ambystoma macrodactylum Baird, 1850)increased use of vegetation cover in lakes with trout and whether adding vegetation structure could reduce predation risk andnonconsumptive effects (NCEs), such as reductions in body size and delayed metamorphosis. We compared use of vegetationcover by larval salamanders in lakes with and without trout and conducted a field experiment to investigate the influence ofadded vegetation structure on salamander body morphology and life history. The probability of catching salamanders in trapsin lakes with trout was positively correlated with the proportion of submerged vegetation and surface cover. Growth rates ofsalamanders in enclosures with trout cues decreased as much as 85% and the probability of metamorphosis decreased by 56%. Wedid not find evidence that adding vegetation reduced NCEs in experimental enclosures, but salamanders in lakes with trout usedmore highly vegetated areas, which suggests that adding vegetation structure at the scale of the whole lake may facilitatecoexistence between salamanders and introduced trout.

Key words: Long-toed Salamander, Ambystoma macrodactylum, experimental enclosures, refuge use, predator cues, submergedvegetation, growth rate.

Résumé : La prédation par des poissons non indigènes a entraîné des réductions d’abondance et l’augmentation du risque dedisparition de populations d’amphibiens a l’échelle planétaire. Bien que rare, la coexistence de poissons et d’amphibiensappétibles a été observée dans des lieux où la végétation aquatique et sa complexité structurale offrent des refuges convenables.Nous avons vérifié si des larves de salamandre a longs doigts (Ambystoma macrodactylum Baird, 1850) utilisaient davantage lecouvert végétal dans des lacs contenant des truites et si l’ajout de structure végétale pouvait réduire le risque de prédation et leseffets non liés a la consommation (ENLC), comme la réduction de la taille du corps et la métamorphose retardée. Nous avonscomparé l’utilisation du couvert végétal par des larves de salamandre dans des lacs avec et sans truites et mené une expériencesur le terrain pour examiner l’influence de l’ajout de structure végétale sur la morphologie du corps et le cycle biologique dessalamandres. La probabilité de capturer des salamandres dans des pièges dans les lacs contenant des truites était positivementcorrélée a la proportion de végétation submergée et de couverture de la surface. Les taux de croissance de salamandres dans desenclos présentant des signaux de truite présentaient des baisses de jusqu’a 85 % et la probabilité de métamorphose diminuait de56 %. Nous n’avons trouvé aucune indication voulant que l’ajout de végétation réduise les ENLC dans les enclos expérimentaux,mais les salamandres dans les lacs contenant des truites utilisaient des zones plus riches en végétation, ce qui donne a penser quel’ajout de structure végétale a l’échelle du lac pourrait faciliter la coexistence de salamandres avec des truites introduites.[Traduit par la Rédaction]

Mots-clés : salamandre a longs doigts, Ambystoma macrodactylum, enclos expérimentaux, utilisation de refuges, signaux deprédateur, végétation submergée, taux de croissance.

IntroductionRefugia are crucial for allowing prey to persist with predators;

complex vegetation structure reduces the maneuverability, visualrange, and effectiveness of predators (Werner et al. 1983; McNair1986; Sih 1987; Kats et al. 1988; Sih et al. 1988). Predator and preypopulations can oscillate in an unstable manner over time, de-pending on the efficiency of a predator at capturing and consum-ing prey (Rosenzweig and MacArthur 1963). However, increases inhabitat complexity reduce the frequency and success of predatorattacks on prey and therefore can stabilize species interactions

(Murdoch and Oaten 1975; Sih 1987; Diehl 1992). Refugia provideprotection from predators, increase food resources for prey, re-duce physical stress and competition among individuals, and sub-sequently, enhance prey density and stable species interactions(Crowder and Cooper 1982; Sih 1987; O’Connor 1991; Diehl 1992).

Trout, salmon, and char (family Salmonidae) have been intro-duced for recreational and commercial purposes worldwide, in-cluding historically fishless waters, resulting in considerableecological effects on native species and ecosystems (Dunham et al.2004; Crawford and Muir 2008; Cucherousset and Olden 2011). Theintroduction of these top predators has resulted in decreases in

Received 10 August 2015. Accepted 28 November 2015.

E.K. Kenison,* A.R. Litt, and T.E. McMahon. Montana State University, Ecology Department, Bozeman, MT 59717, USA.D.S. Pilliod. U.S. Geological Survey, Forest and Rangeland Ecosystem Science Center, Boise, ID 83706, USA.Corresponding author: Erin K. Kenison (emails: ekenison@purdue.edu, erin.kenison@gmail.com).*Present address: Purdue University, West Lafayette, IN 47907, USA.

243

Can. J. Zool. 94: 243–249 (2016) dx.doi.org/10.1139/cjz-2015-0160 Published at www.nrcresearchpress.com/cjz on 21 January 2016.

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Purd

ue U

niv.

Lib

. TSS

on

03/2

3/16

For

pers

onal

use

onl

y.

abundance and extirpations of amphibian populations (Katsand Ferrer 2003; Pilliod et al. 2012). The Long-toed Salamander(Ambystoma macrodactylum Baird, 1850), one of the most widelydistributed amphibians west of the Continental Divide, is espe-cially vulnerable to nonnative, introduced fish. This salamander ispalatable, lacks chemical defenses, and has an extended larvalperiod (1–3 years), during which larvae must overwinter in deep,ice-free lakes often alongside predatory fish (Welsh et al. 2006;Pilliod et al. 2010). In lakes with trout, the abundance of salaman-der egg masses is lower, larvae are >65% less abundant, and somepopulations have been extirpated (Tyler et al. 1998a; Funk andDunlap 1999; Pilliod and Peterson 2001; Bull and Marx 2002;Welsh et al. 2006; Hirner and Cox 2007; Pearson and Goater 2008;Pilliod et al. 2010).

The few lakes where salamanders and trout have been found toco-occur (i.e., existing in the same lake together) have emergentvegetation and shallow littoral zones that likely provide adequaterefugia from fish predators (Tyler et al. 1998a; Pearson and Goater2008; Pilliod et al. 2010, 2013). Salamanders are capable of detect-ing predator cues and respond by increasing refuge use to avoidpredation, but incur concomitant reductions in energy uptake(Semlitsch 1987; Figiel and Semlitsch 1990; Skelly and Werner1990; Tyler et al. 1998b; Kenison et al. 2016). Salamanders thatco-occur with trout exhibit a number of nonconsumptive effects(NCEs; Davenport et al. 2014), including reduction in size andgrowth and delayed metamorphosis, which are induced by thepresence of trout or their cues alone (Kenison et al. 2016). Suchstrong effects on body morphology and life history in the pres-ence of trout are likely to have significant effects on predation riskand reproductive success (Semlitsch 1987).

The objectives of our study were to (i) examine refuge use bylarval Long-toed Salamanders in lakes with and without trout and(ii) test whether the addition of complex vegetation structure canreduce the NCEs of trout on salamander larvae and thereby facili-tate persistence of salamanders in lakes where nonnative trout arepresent. We predicted that salamanders in lakes with trout would befound in denser cover compared with lakes without trout. We alsopredicted that individuals would be smaller and less likely tometamorphose in enclosures with trout cues than in enclosureswithout trout cues, but that added vegetation structure wouldreduce these NCEs, such that salamanders in enclosures with andwithout trout cues would be comparable in size. Our study is thefirst to experimentally supplement available refugia in an at-tempt to reduce NCEs of predatory trout on salamanders andprovide a more feasible management strategy to trout removal.

Materials and methods

Salamander refuge useWe compared refuge use of Long-toed Salamanders in 14 per-

manent lakes, 7 lakes with trout and 7 lakes without trout, in theFlathead National Forest between the Swan and the MissionMountain ranges in northwestern Montana during the summer of2012 (Kenison et al. 2016). Trout species in our study lakes includedrainbow trout (Oncorhynchus mykiss (Walbaum, 1792)), cutthroattrout (Oncorhynchus clarkii (Richardson, 1836)), or hybrids of thetwo; we were unable to determine the density of fish in each studylake due to time constraints and equipment limitations. We cap-tured salamander larvae with mesh, collapsible minnow traps(0.25 m long × 0.25 m wide × 0.43 m high), which are associatedwith little mortality or injury and are effective at capturing sala-manders passively without bait (Adams et al. 1997). We set alltraps in the littoral zone around the perimeter of each lake atdepths ≤1 m. The total number of traps in each lake depended onlake size and ranged from 6 to 16; lake perimeters ranged between110 and 960 m. We deployed traps in the same locations in eachlake for four, 4-day trapping sessions from July through August2012.

We quantified submerged vegetation and surface cover at everyother trap location in all study lakes during each of the fourtrapping sessions as measures of habitat complexity or refugia.We quantified the proportion of submerged vegetation by restinga 1 m long chain, composed of 44 links, each 2.3 cm long, along thelake bottom at every other trap location. We counted the numberof chain links in contact with vegetation under the water (i.e.,living material such as stems) and converted counts of chain linksinto proportion of submerged vegetation (Kinsolving and Bain1990). We also estimated the proportion of emergent vegetationon the water surface (i.e., surface cover) within a 1 m2 quadratabove every other trap location. We measured habitat character-istics at every other trap location rather than every location be-cause this design allowed us a reasonable, yet more feasiblemeasure of habitat along the entire lake perimeter.

Enclosure experimentIn 2013, we conducted a field experiment in 5 of the original

14 lakes, 2 lakes with trout and 3 lakes without trout, to investi-gate the effectiveness of adding vegetation structure to mitigateNCEs of trout on larval salamanders. The five study lakes selectedhad the highest densities of salamanders, which ensured that wehad sufficient individuals for experiments without the need forbetween-lake transplants. We constructed experimental field en-closures with PVC pipe and fiberglass window screening (2 m long ×1 m wide × 1 m high) (Sredl and Collins 1991; Kiesecker and Blaustein1998). PVC pipes provided structure and support for the rectangu-lar frame and screening was secured around the four sides and thebottom of the enclosure to allow continuous water exchange be-tween enclosures and the lake. We covered the top of the enclo-sure with chicken wire to restrict avian predators. Experimentaltreatments were based on a 2 × 2 factorial arrangement: (i) pres-ence or absence of trout cues (i.e., enclosures in lakes with troutcompared with those without trout) and (ii) presence or absence ofadded vegetation structure. The trout treatments exposed larvalsalamanders to the chemical or visual signals of predators, butprotected larvae from direct predation. This factorial design al-lowed us to test for the effects of refuge, the effects of trout cues,and the combined effects of trout cues and added vegetation.

We constructed complex vegetation structures by binding ap-proximately 20 dead, dry sticks with 20 living, leafy branches (i.e.,aspen or other littoral vegetation) with twine in an inverted fun-nel shape (Fig. 1; Schneider and Winemiller 2008). All vegetationwas collected at study sites and not transported between lakes.Vegetation structures were roughly 50 cm in circumference at thebase and 50–60 cm tall. We placed two vegetation structures ineach enclosure that was assigned to added vegetation treatments,one structure at each end of the enclosure, and secured them inplace with a large rock.

We placed enclosures in the littoral zone, along the north-facing edge of each lake to minimize potential differences amongsites (e.g., temperature) within a lake. We positioned enclosuresso that depth, distance between enclosures, and distance fromshore were equal for all enclosures (Sredl and Collins 1991). Wealso added �10 L of lake substrate to the bottom of each enclosureto incorporate natural sediment. Lakes with trout had four repli-cates of each of the two possible treatments (presence of troutcues, with and without added vegetation structure) for a total ofeight enclosures per lake. Lakes without trout had two replicatesof each of the two possible treatments (absence of trout cues, withand without added vegetation structure) for a total of four enclo-sures per lake.

We captured salamanders with minnow traps and added indi-viduals that were comparable in size (>0.2 g) to experimentalenclosures during June and July 2013. We anesthetized salaman-der larvae with MS-222, measured mass, snout–vent length (SVL),total length, tail length, and tail depth (height of the tail) with anelectronic scale (to the nearest 0.001 g) and calipers (to the nearest

244 Can. J. Zool. Vol. 94, 2016

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Purd

ue U

niv.

Lib

. TSS

on

03/2

3/16

For

pers

onal

use

onl

y.

0.1 mm), and marked each individual using visual implant elasto-mer with a unique two-color mark in the middle of their tail. Werandomly assigned 20 salamanders to each enclosure; salamanderdensities were comparable with previous studies that did not ob-serve cannibalism and where adequate amounts of food persistedthroughout the study periods (e.g., Clark 1986; Semlitsch 1987; Tyleret al. 1998b; Pearson and Goater 2009). Although we expected zoo-plankton and other small crustaceans to be able to pass throughthe enclosure screen, we added �10 L of lake water to enclosuresat each visit to provide additional food resources (Stenhouse 1985;Tarr and Babbitt 2002).

After adding salamanders, we visited each enclosure four addi-tional times during July and August 2013. At each enclosure visit,we captured individual larvae with hand nets and recorded mea-sures of body morphology (i.e., mass, SVL, total length, tail length,and tail depth). We also noted whether salamanders had initiatedmetamorphosis, which we defined as any evidence of gill absorp-tion (Dodd and Dodd 1976; Arntzen 1981; Duellman and Trueb1986). After we completed data collection, we released individualsthat had absorbed their gills completely into the lake and re-turned other individuals to assigned enclosures. We were unableto measure survival accurately, as we could not distinguish withcertainty whether salamanders missing from enclosures escapedor died due to predation by invertebrates, conspecific cannibal-ism, or natural causes. At the end of the experimental period,salamander densities were comparable between enclosures inlakes with and without trout. We terminated the experiment atthe end of August, released all remaining individuals in the lake,and removed enclosures. We anesthetized, marked, and handledall captured individuals in accordance with Montana State Univer-sity Institutional Animal Care and Use Committee protocols2012-28 and 2013-04.

Statistical analysesWe assessed refuge use by quantifying the relationship between

the number of salamanders captured and the proportion of sub-merged vegetation and surface cover; we compared this relation-ship between lakes with and without trout. We assessed theimportance of adding vegetation structures by comparing sala-

mander growth rates, size of salamanders at metamorphosis, andthe probability of metamorphosis across all four experimentaltreatments: (1) presence of trout cues (i.e., lakes with trout) andadded structure, (2) presence of trout cues and no added structure,(3) absence of trout cues (i.e., lakes without trout) and added struc-ture, and (4) absence of trout cues and no added structure. Weused generalized linear mixed models with random slopes andintercepts for all of our analyses to account for repeated measure-ments in lakes and over time (Zuur et al. 2009).

To assess refuge use, we included the presence–absence of troutand vegetation measurements as main effects in our analyticalmodels and also included the date of each trapping session toaccount for changes in vegetation abundance over time. Wetested for interactive effects between presence–absence of troutand vegetation measurements, but removed interactions that didnot explain sufficient variation (P > 0.1). We treated lakes as sub-jects and used a nested data structure that included individualtraps nested within lakes, sampled over multiple visits, as ourrandom effects. We excluded two lakes from these analyses be-cause we did not capture any salamanders in traps where vegeta-tion was measured.

To assess the importance of adding vegetation structures on theNCEs of trout in our enclosure experiment, we considered allmeasurements of body morphology (i.e., mass, SVL, total length,tail length, and tail depth) as separate response variables in ourmodels. We included experimental treatments (i.e., presence–absence of trout cues (trout) and presence–absence of vegetationstructure (structure)) as main effects. We quantified growth ratesand size at metamorphosis as a function of trout cues and vegeta-tion structure by examining simple (trout, structure) and interac-tive effects (trout × structure). We also considered all possibletwo-way interactions between treatment factors and time (trout ×structure, trout × day, structure × day) in models for probability ofmetamorphosis, but removed interactions that did not explainsufficient variation (P > 0.1). We included a nested data structurethat matched our design; we collected multiple measurements ofsalamanders within enclosures within the study lakes, sampledover time (day: modeled as the number of days since the firstenclosure visit), as random effects.

We ran all statistical analyses in program R version 3.0.2 andused the nlme and MASS packages (R Core Team 2013). We reportprobability of salamander capture as a function of the proportionof submerged vegetation and surface cover. We estimate differ-ences in growth rates, size at metamorphosis, and the probabilityof metamorphosis for salamanders in enclosures. We also provide95% confidence intervals (CI) for all estimates.

Results

Salamander refuge useThe probability of salamander capture was related to the pro-

portion of submerged vegetation, but the magnitude and direc-tion of this relationship depended on presence–absence of trout(Fig. 2A). In lakes with trout, salamanders were more likely to becaptured in areas with more submerged vegetation, whereas thisrelationship reversed in the absence of trout (Fig. 2A). For eachadditional 0.1 increase in the proportion of submerged vegeta-tion, the odds of capture increased by a factor of 1.25 (95% CI =0.75–2.07) in lakes with trout or by a factor of 0.82 (0.66–1.01) inlakes without trout (t[465] = 2.86, P = 0.004; Fig. 2A). We found someevidence that salamanders preferred areas with more vegetationcover at the water surface, but this relationship did not depend onthe presence–absence of trout (t[466] = 1.54, P = 0.123; Fig. 2B). Forevery 0.1 increase in the proportion of surface cover, the odds ofcapture increased by a factor of 1.17 (0.65–1.44), regardless ofwhether trout were present.

Fig. 1. Vegetation structure constructed for enclosures made withdead and living plant materials and secured with twine.

Kenison et al. 245

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Purd

ue U

niv.

Lib

. TSS

on

03/2

3/16

For

pers

onal

use

onl

y.

Enclosure experimentAdding vegetation structure to experimental field enclosures

had no detectable effect on growth rates or body morphology ofsalamanders, but we found substantial differences among bodymeasurements between enclosures with and without trout cues(Fig. 3, Table 1). Salamanders in enclosures without trout cues, onaverage, gained 64% more mass (95% CI = 48%–90%) and grew 54.5%more inSVL (43%–70%), 58%more intotal length (45%–77%), 57%more intail length (42%–82%), and 85% more in tail depth (77%–92%) per daycompared with salamanders in enclosures with trout cues.

Body morphology at the time of metamorphosis was similaracross all treatments (Table 2), but the probability of metamor-phosing depended on presence–absence of trout. Salamanders ex-

posed to trout cues had the lowest probability of metamorphosisthroughout the experiment, compared with salamanders not ex-posed to trout cues, regardless of the presence of added vegetationstructure (Fig. 4). On average, for each additional day, the odds thatsalamanders in enclosures without trout cues would metamorphosewas 15% higher (95% CI = 9%–23%) compared with the odds of sala-manders metamorphosing in enclosures with trout cues, after ac-counting for vegetation treatment (Fig. 4; t[1916] = –3.31, P = 0.001).

DiscussionIn our study, the greater use of areas with more submerged

vegetation by Long-toed Salamanders in lakes with trout supports

Fig. 2. Probability of capturing Long-toed Salamanders (Ambystoma macrodactylum) as a function of (A) the proportion of submerged vegetationand (B) the proportion of vegetation cover on the water surface (n = 398 salamanders in 67 traps) in northwestern Montana during thesummer of 2012. Points are jittered to better distinguish salamander capture data.

Fig. 3. Rates of growth per day (with 95% CIs) for Long-toed Salamanders (Ambystoma macrodactylum) in enclosures without trout cues orstructure (– trout, – structure; n = 120 salamanders, 6 enclosures), without trout cues but with added structure (– trout, + structure;n = 123 salamanders, 6 enclosures), with trout cues but without structure (+ trout, – structure; n = 160 salamanders, 8 enclosures), and withtrout cues and added structure (+ trout, + structure; n = 160 salamanders, 8 enclosures) in northwestern Montana during the summer of 2013.

246 Can. J. Zool. Vol. 94, 2016

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Purd

ue U

niv.

Lib

. TSS

on

03/2

3/16

For

pers

onal

use

onl

y.

the importance of refugia as protection from predators and thepotential role in promoting coexistence between aquatic preda-tors and prey in lentic systems. These findings are congruent withprevious studies which found that species richness and occupancyof amphibians within lakes with predatory fish are positively cor-related with emergent vegetation (Hecnar and M’Closkey 1997;Hartel et al. 2007). A study of 1749 water bodies in the northernRocky Mountains found that Long-toed Salamanders only coex-isted with trout in 47 (2.7%) lakes, yet 77% of the lakes with coex-isting trout and salamanders contained highly vegetated shallows(76%–100% cover; Pilliod et al. 2013). In experiments with free-swimming predators, survival of larval amphibians increases inareas with higher cover and density of plants (Babbitt and Tanner1997; Baber and Babbitt 2004). The use of refugia by prey can

reduce a predator’s ability to detect and capture prey, increasingthe potential for predator–prey coexistence (Murdoch and Oaten1975; Diehl 1992; Bartholomew et al. 2000).

Larval amphibians that are capable of detecting the physicalpresence or chemical cues of predators often increase refuge useto avoid predation (Petranka et al. 1987; Kats et al. 1988; Davenportet al. 2014). For example, tadpoles subjected to a caged predator orfish chemical cues may reduce the time spent foraging by as muchas 41% and reduce the time spent outside of refugia by 68%(Petranka et al. 1987; Skelly and Werner 1990). Although refugeuse reduces the risk of consumption, prey can incur NCEs due tochanges in foraging, allocating resources to alter morphology andproduce chemical defenses, or both (Brönmark and Miner 1992;Hagman et al. 2009). Long-toed Salamanders that co-occur in lakeswith trout weigh 38% less and are 24% shorter compared withsalamanders in lakes without fish predators (Kenison et al. 2016).Persisting in environments with fish that are perceived to be highrisk also increases stress, which can inhibit food intake, suppressappetite, and further contribute to NCEs (Crespi and Denver2005). Our study demonstrates that salamanders are able to ap-propriately detect predators and employ avoidance behaviors, butface a trade-off between foraging and safety, given that salaman-ders exposed to trout cues in our experimental enclosures exhib-ited slower growth rates and delayed metamorphosis.

Salamanders must reach a minimum body size before meta-morphosis can begin (Wilbur and Collins 1973; Nicieza 2000;Babbitt 2001; Altwegg 2002; Benard 2004). We found the probabil-ity of metamorphosis was lower in enclosures with trout cues,which suggests that salamanders decrease activity in the presenceof predators, reduce growth rates, and subsequently, extend lar-val periods to reach necessary size thresholds. This finding is con-gruent with studies of other amphibian species: Southern LeopardFrogs (Lithobates sphenocephalus (Cope, 1886)), Wood Frogs (Lithobatessylvaticus (LeConte, 1825)), and Water Frogs (Pelophylax ridibundus(Pallas, 1771)) extend their larval period when exposed to insectpredators, but metamorphose at similar sizes (Babbitt 2001;Relyea 2001; Van Buskirk and Saxer 2001). Similarly, Long-toedSalamanders have reduced growth rates and prolonged larval pe-riods in the presence of cannibalistic conspecifics, but are similarin size at metamorphosis (Wildy et al. 1999). Although uncom-

Table 1. Factors affecting growth rate (change in size over time, denoted by “date”) of Long-toed Salamanders (Ambystoma macrodactylum) inexperimental enclosures without trout cues or structure (no trout, no structure; n = 120 salamanders, 6 enclosures), without trout cues but withadded structure (no trout, added structure; n = 123 salamanders, 6 enclosures), with trout cues but without structure (trout, no structure; n =160 salamanders, 8 enclosures), and with trout cues and added structure (trout, added structure; n = 160 salamanders, 8 enclosures) in north-western Montana during the summer of 2013.

Mass (g) SVL (mm) Total length (mm) Tail length (mm) Tail depth (mm)

Explanatory df Estimate t P Estimate t P Estimate t P Estimate t P Estimate t P

Trout 3 0.34 0.97 0.402 4.10 1.02 0.381 8.09 1.16 0.331 3.84 1.05 0.371 0.64 0.73 0.518Structure 22 0.01 0.42 0.678 0.34 0.78 0.446 0.84 0.88 0.389 0.05 0.11 0.912 0.12 0.96 0.347Date 1271 0.03 8.56 <0.001 0.37 11.19 <0.001 0.69 10.33 <0.001 0.31 7.73 <0.001 0.07 18.23 <0.001Trout × date 1271 −0.02 −3.42 <0.001 −0.19 −3.35 <0.001 −0.39 −3.79 <0.001 −0.18 −2.87 0.004 −0.05 −13.15 <0.001Structure × date 1271 −0.00 −2.08 0.037 −0.04 −2.34 0.020 −0.58 −2.08 0.038 — — — −0.01 −3.31 0.001

Note: We removed interactions that did not explain sufficient variation. The variability in degrees of freedom (df) is a function of accounting for the nested studydesign. SVL, snout–vent length.

Table 2. Factors affecting body morphology of Long-toed Salamanders (Ambystoma macrodactylum) at metamorphosis in experimental enclosureswithout trout cues or structure (n = 19 salamanders, 6 enclosures), without trout cues but with added structure (n = 24 salamanders, 6 enclosures),with trout cues but without structure (n = 52 salamanders, 8 enclosures), and with trout cues and added structure (n = 35 salamanders,8 enclosures), after accounting for time, in northwestern Montana during the summer of 2013.

Mass (g) SVL (mm) Total length (mm) Tail length (mm) Tail depth (mm)

Explanatory df Difference t P Difference t P Difference t P Difference t P Difference t P

Trout 3 −0.19 −0.65 0.563 −0.45 −0.36 0.740 −3.46 −0.79 0.488 −2.13 −0.90 0.432 −0.30 −0.54 0.624Structure 18 −0.01 −0.19 0.855 0.16 0.45 0.659 −0.39 −0.36 0.720 −0.67 −0.71 0.486 −0.44 −1.31 0.206

Note: Estimated differences compare measurements against the treatment without trout cues and structure. SVL, snout–vent length.

Fig. 4. Probability of metamorphosis over time for Long-toedSalamanders (Ambystoma macrodactylum) in enclosures without troutcues or added structure (– trout, – structure; n = 19 of 120 salamanders),without trout cues but with added structure (– trout, + structure;n = 24 of 123 salamanders), with trout cues but without addedstructure (+ trout, – structure; n = 36 of 160 salamanders), and withtrout cues and added structure (+ trout, + structure; n = 47 of160 salamanders) in northwestern Montana. Day 0 is 30 June andday 43 is 12 August during the summer of 2013.

Kenison et al. 247

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Purd

ue U

niv.

Lib

. TSS

on

03/2

3/16

For

pers

onal

use

onl

y.

mon, some species are able to metamorphose early when sub-jected to increased risk of predation. For example, Western Toads(Anaxyrus boreas (Baird and Girard, 1852); formerly Bufo boreas Bairdand Girard, 1852) decrease time to metamorphosis by �10 dayswhen exposed to predatory cues of backswimmers (species of thegenus Notonecta L., 1758; Chivers et al. 1999; Relyea 2007).

Fish introductions have been associated with reduced popula-tions of macroinvertebrates, zooplankton, and subsequently,amphibians (Knapp et al. 2001). Given that trout and larvalsalamanders have overlapping diets, changes in food availabilityin lakes with trout may have influenced salamander growth andsize. However, resource competition is inherently related to non-consumptive effects of trout. Although we do not have data onfish densities or resource availability for our study lakes, we pres-ent strong evidence for a change in growth and probability ofmetamorphosis as a function of trout cues. Salamanders in enclo-sures were protected from potential consumptive effects of troutpredators, yet they still demonstrated changes in growth and lifehistory. We argue that changes incurred by salamanders, regard-less of the exact mechanism (e.g., altered foraging behavior orresource competition), still provide evidence of nonconsumptiveeffects of trout.

Salamanders that extend their larval period in lakes with troutmay increase their exposure to predators and risk of mortality,thus reducing the number of individuals that leave the pond tosuccessfully enter the terrestrial system as adults. Although wedid not find evidence that adding vegetation structure helped tomitigate NCEs of trout, Orizaola and Braña (2003) found that moreabundant available refugia increased survival of larval newtsmore than two-fold in the presence of fish predators. We foundthat salamanders in lakes with trout used more densely vegetatedareas; we captured salamanders more often in open water in lakeswithout trout. Dense vegetation may increase survival in thepresence of trout, but may also provide less-than-ideal foraging areasfor salamanders. In the presence of piscivorous largemouth bass(Micropterus salmoides (Lacepède, 1802)), small bluegills (Lepomismacrochirus Rafinesque, 1819) survive by foraging in protected veg-etated habitat, but only acquire one-third of the resources foundin open water (Werner et al. 1983). Habitat features such as rocks,woody material, vegetation, and small interstitial spaces provideeffective protection from fish predators and also are associatedwith increased density and species richness of invertebrates(Stenhouse 1985; Babbitt and Jordan 1996; Hartel et al. 2007). Ourexperimental vegetation structures may have been too small toreduce the perceived risk of predation and may not have been inplace long enough to provide sufficient food resources for larvalsalamanders. Therefore, by adding larger, more diverse structuresto lakes and allowing time for invertebrate prey to establish, wemight be able to create highly productive, food-rich areas, wheresalamanders can forage efficiently and safely from trout.

Complete removal of fish from mountain lakes has been suc-cessful in increasing population sizes of amphibians (Knapp et al.2007). However, future research focused on adding complex struc-ture at a whole-lake scale will be important to understand whetherthis strategy can be effective at reducing the consumptive andnonconsumptive effects of trout while preserving recreationalfishing opportunities. Previous conservation projects for amphib-ians have been limited to maintaining buffer zones around breedingsites, creating connectivity between wetlands, or constructingnew and restoring anthropogenically altered water bodies (Semlitsch2002; Shulse et al. 2012). If adding vegetation structure at largerscales can increase resource availability for larval salamanders,then this management strategy may be the most feasible to miti-gate NCEs of nonnative trout without requiring the removal offish.

AcknowledgementsThis work was funded by Montana State University, with addi-

tional financial support from Montana Fish, Wildlife and Parks,Counter Assault, and Montana Institute on Ecosystems. We aregrateful to M. Forzley for his work as a field technician, B. Tornabenefor his statistical assistance, and reviewers for their constructivecomments. Any use of trade, product, or firm names is for descrip-tive purposes only and does not imply endorsement by the U.S.Government.

ReferencesAdams, M.J., Richter, K.O., and Leonard, W.P. 1997. Surveying and monitoring

amphibians using aquatic funnel traps. Northwest Fauna, 4: 47–54.Altwegg, R. 2002. Predator-induced life-history plasticity under time constraints

in pool frogs. Ecology, 83(9): 2542–2551. doi:10.1890/0012-9658(2002)083[2542:PILHPU]2.0.CO;2.

Arntzen, J. 1981. Ecological observations on Chioglossa lusitanica (Caudate, Sala-mandridae). Amphib.-Reptilia, 1(3–4): 187–204. doi:10.1163/156853881X00311.

Babbitt, K.J. 2001. Behaviour and growth of southern leopard frog (Ranasphenocephala) tadpoles: effects of food and predation risk. Can. J. Zool. 79(5):809–814. doi:10.1139/z01-040.

Babbitt, K.J., and Jordan, F. 1996. Predation on Bufo terrestris tadpoles: effects ofcover and predator identity. Copeia, 1996(2): 485–488. doi:10.2307/1446873.

Babbitt, K.J., and Tanner, G.W. 1997. Effects of cover and predator identity onpredation of Hyla squirella tadpoles. J. Herpetol. 31(1): 128–130. doi:10.2307/1565342.

Baber, M.J., and Babbitt, K.J. 2004. Influence of habitat complexity on predator–prey interactions between the fish (Gambusia holbrooki) and tadpoles of Hylasquirella and Gastrophryne carolinensis. Copeia, 2004(1): 173–177. doi:10.1643/CE-03-056R1.

Bartholomew, A., Diaz, R.J., and Cicchetti, G. 2000. New dimensionless indices ofstructural habitat complexity: predicted and actual effects on a predator’sforaging success. Mar. Ecol. Prog. Ser. 206: 45–58. doi:10.3354/meps206045.

Benard, M.F. 2004. Predator-induced phenotypic plasticity in organisms withcomplex life histories. Annu. Rev. Ecol. Evol. Syst. 35(1): 651–673. doi:10.1146/annurev.ecolsys.35.021004.112426.

Brönmark, C., and Miner, J.G. 1992. Predator-induced phenotypical change inbody morphology in crucian carp. Science, 258(5086): 1348–1350. doi:10.1126/science.258.5086.1348. PMID:17778362.

Bull, E.L., and Marx, D.B. 2002. Influence of fish and habitat on amphibiancommunities in high elevation lakes in northeastern Oregon. Northwest Sci.76(3): 240–248.

Chivers, D.P., Kiesecker, J.M., Marco, A., Wildy, E.L., and Blaustein, A.R. 1999.Shifts in life history as a response to predation in western toads (Bufo boreas).J. Chem. Ecol. 25(11): 2455–2463. doi:10.1023/A:1020818006898.

Clark, K.L. 1986. Responses of spotted salamander, Ambystoma maculatum, popu-lations in central Ontario to habitat acidity. Can. Field-Nat. 100(4): 463–469.

Crawford, S.S., and Muir, A.M. 2008. Global introductions of salmon and trout inthe genus Oncorhynchus: 1870–2007. Rev. Fish Biol. Fishes, 18: 313–344. doi:10.1007/s11160-007-9079-1.

Crespi, E.J., and Denver, R.J. 2005. Roles of stress hormones in food intakeregulation in anuran amphibians throughout the life cycle. Comp. Biochem.Physiol. Part A Mol. Integr. Physiol. 141(2005): 381–390. doi:10.1016/j.cbpb.2004.12.007.

Crowder, L.B., and Cooper, W.E. 1982. Habitat structural complexity and theinteraction between bluegills and their prey. Ecology, 63(6): 1802–1813. doi:10.2307/1940122.

Cucherousset, J., and Olden, J.D. 2011. Ecological impacts of nonnative freshwa-ter fishes. Fisheries (Bethesda), 36(5): 215–230. doi:10.1080/03632415.2011.574578.

Davenport, J.M., Hossack, B.R., and Lowe, W.H. 2014. Partitioning the non-consumptive effects of predators on prey with complex life histories. Oecologia,176: 149–155. doi:10.1007/s00442-014-2996-5. PMID:24965938.

Diehl, S. 1992. Fish predation and benthic community structure: the role ofomnivory and habitat complexity. Ecology, 73(5): 1646–1661. doi:10.2307/1940017.

Dodd, M.H.I., and Dodd, J.M. 1976. The biology of metamorphosis. In Physiologyof the Amphibia. Edited by B. Lofts. Academic Press, New York. pp. 467–599.

Duellman, W.E., and Trueb, L. 1986. Biology of amphibians. McGraw-Hill, NewYork.

Dunham, J.B., Pilliod, D.S., and Young, M.K. 2004. Assessing the consequences ofnonnative trout in headwater ecosystems in western North America. Fisheries(Bethesda), 29(6): 18–26. doi:10.1577/1548-8446(2004)29[18:ATCONT]2.0.CO;2.

Figiel, C.R., Jr., and Semlitsch, R.D. 1990. Population variation in survival andmetamorphosis of larval salamanders (Ambystoma maculatum) in the presenceand absence of fish predation. Copeia, 1990(3): 818–826. doi:10.2307/1446447.

Funk, W.C., and Dunlap, W.W. 1999. Colonization of high-elevation lakes bylong-toed salamanders (Ambystoma macrodactylum) after the extinction of in-troduced trout populations. Can. J. Zool. 77(11): 1759–1767. doi:10.1139/z99-160.

248 Can. J. Zool. Vol. 94, 2016

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Purd

ue U

niv.

Lib

. TSS

on

03/2

3/16

For

pers

onal

use

onl

y.

Hagman, M., Hayes, R.A., Capon, R.J., and Shine, R. 2009. Alarm cues experi-enced by cane toad tadpoles affect post-metamorphic morphology and chem-ical defences. Funct. Ecol. 23: 126–132. doi:10.1111/j.1365-2435.2008.01470.x.

Hartel, T., Nemes, S., Cogalniceanu, D., Öllerer, K., Schweiger, O., Moga, C., andDemeter, L. 2007. The effect of fish and aquatic habitat complexity on am-phibians. Hydrobiologia, 583: 173–182. doi:10.1007/s10750-006-0490-8.

Hecnar, S.J., and M’Closkey, R.T. 1997. The effects of predatory fish on amphibianspecies richness and distribution. Biol. Conserv. 79: 123–131. doi:10.1016/S0006-3207(96)00113-9.

Hirner, J.L.M., and Cox, S.P. 2007. Effects of rainbow trout (Oncorhynchus mykiss)on amphibians in productive recreational fishing lakes of British Columbia.Can. J. Fish. Aquat. Sci. 64(12): 1770–1780. doi:10.1139/f07-139.

Kats, L.B., and Ferrer, R.P. 2003. Alien predators and amphibian declines: reviewof two decades of science and the transition to conservation. Divers. Distrib.9: 99–110. doi:10.1046/j.1472-4642.2003.00013.x.

Kats, L.B., Petranka, J.W., and Sih, A. 1988. Antipredator defenses and the per-sistence of amphibian larvae with fishes. Ecology, 69(6): 1865–1870. doi:10.2307/1941163.

Kenison, E.K., Litt, A.R., Pilliod, D.S., and McMahon, T. 2016. Larval long-toedsalamanders incur nonconsumptive effects in the presence of nonnativetrout. Ecosphere. [In press.]

Kiesecker, J.M., and Blaustein, A.R. 1998. Effects of introduced bullfrogs andsmallmouth bass on microhabitat use, growth, and survival of native red-legged frogs (Rana aurora). Conserv. Biol. 12(4): 776–787. doi:10.1046/j.1523-1739.1998.97125.x.

Kinsolving, A.D., and Bain, M.B. 1990. A new approach for measuring cover infish habitat studies. J. Freshw. Ecol. 5(3): 373–378. doi:10.1080/02705060.1990.9665250.

Knapp, R.A., Matthews, K.R., and Sarnelle, O. 2001. Resistance and resilience ofalpine lake fauna to fish introductions. Ecol. Monogr. 71: 401–421. doi:10.1890/0012-9615(2001)071[0401:RAROAL]2.0.CO;2.

Knapp, R.A., Boiano, D.M., and Vredenburg, V.T. 2007. Removal of nonnative fishresults in population expansion of a declining amphibian (mountain yellow-legged frog, Rana muscosa). Biol. Conserv. 135: 11–20. doi:10.1016/j.biocon.2006.09.013. PMID:17396156.

McNair, J.N. 1986. The effects of refuges on predator–prey interactions: a recon-sideration. Theor. Popul. Biol. 29: 38–63. doi:10.1016/0040-5809(86)90004-3.PMID:3961711.

Murdoch, W.W., and Oaten, A. 1975. Predation and population stability. Adv.Ecol. Res. 9: 1–129. doi:10.1016/S0065-2504(08)60288-3.

Nicieza, A.G. 2000. Interacting effects of predation risk and food availability onlarval anuran behaviour and development. Oecologia, 123: 497–505. doi:10.1007/s004420000343.

O’Connor, N.A. 1991. The effects of habitat complexity on the macroinverte-brates colonizing wood substrates in a lowland stream. Oecologia, 85: 504–512. doi:10.1007/BF00323762.

Orizaola, G., and Braña, F. 2003. Response of predator-naïve newt larvae to foodand predator presence. Can. J. Zool. 81(11): 1845–1850. doi:10.1139/z03-160.

Pearson, K.J., and Goater, C.P. 2008. Distribution of long-toed salamanders andintroduced trout in high- and low-elevation wetlands in southwestern Al-berta, Canada. Ecoscience, 15: 453–459. doi:10.2980/15-4-3127.

Pearson, K.J., and Goater, C.P. 2009. Effects of predaceous and nonpredaceousintroduced fish on the survival, growth, and antipredation behaviours oflong-toed salamanders. Can. J. Zool. 87(10): 948–955. doi:10.1139/Z09-084.

Petranka, J.W., Kats, L.B., and Sih, A. 1987. Predator–prey interactions amongfish and larval amphibians: use of chemical cues to detect predatory fish.Anim. Behav. 35: 420–425. doi:10.1016/S0003-3472(87)80266-X.

Pilliod, D.S., and Peterson, C.R. 2001. Local and landscape effects of introducedtrout on amphibians in historically fishless watersheds. Ecosystems, 4: 322–333. doi:10.1007/s10021-001-0014-3.

Pilliod, D.S., Hossack, B.R., Bahls, P.F., Bull, E.L., Corn, P.S., Hokit, G.,Maxell, B.A., Munger, J.C., and Wyrick, A. 2010. Non-native salmonids affectamphibian occupancy at multiple spatial scales. Divers. Distrib. 16: 959–974.doi:10.1111/j.1472-4642.2010.00699.x.

Pilliod, D.S., Griffiths, R.A., and Kuzmin, S.L. 2012. Ecological impacts of non-native species. In Conservation and decline of amphibians: ecological as-pects, effect of humans, and management. Vol. 10. Edited by H. Heatwole andJ.W. Wilkinson. Surrey Beatty and Sons, NSW, Australia. pp. 3343–3382.

Pilliod, D.S., Arkle, R.S., and Maxell, B.A. 2013. Persistence and extirpation in

invaded landscapes: patch characteristics and connectivity determine effectsof non-native predatory fish on native salamanders. Biol. Invasions, 15: 671–685. doi:10.1007/s10530-012-0317-7.

R Core Team. 2013. R: a language and environment for statistical computing.R Foundation for Statistical Computing, Vienna, Austria. Available fromhttps://www.r-project.org/.

Relyea, R.A. 2001. Morphological and behavioral plasticity of larval anurans inresponse to different predators. Ecology, 82(2): 523–540. doi:10.1890/0012-9658(2001)082[0523:MABPOL]2.0.CO;2.

Relyea, R.A. 2007. Getting out alive: how predators affect the decision to meta-morphose. Oecologia, 152(3): 389–400. doi:10.1007/s00442-007-0675-5. PMID:17356812.

Rosenzweig, M.L., and MacArthur, R.H. 1963. Graphical representation and sta-bility conditions of predator–prey interactions. Am. Nat. 97(895): 209–223.doi:10.1086/282272.

Schneider, K.N., and Winemiller, K.O. 2008. Structural complexity of woodydebris patches influences fish and macroinvertebrate species richness in atemperate floodplain-river system. Hydrobiologia, 610: 235–244. doi:10.1007/s10750-008-9438-5.

Semlitsch, R.D. 1987. Interactions between fish and salamander larvae. Costs ofpredator avoidance or competition? Oecologia, 72(4): 481–486. doi:10.1007/BF00378972.

Semlitsch, R.D. 2002. Critical elements for biologically based recovery plans ofaquatic-breeding amphibians. Conserv. Biol. 16(3): 619–629. doi:10.1046/j.1523-1739.2002.00512.x.

Shulse, C.D., Semlitsch, R.D., Trauth, K.M., and Gardner, J.E. 2012. Testing wet-land features to increase amphibian reproductive success and species rich-ness for mitigation and restoration. Ecol. Appl. 22(5): 1675–1688. doi:10.1890/11-0212.1. PMID:22908722.

Sih, A. 1987. Prey refuges and predator–prey stability. Theor. Popul. Biol. 31: 1–12.doi:10.1016/0040-5809(87)90019-0.

Sih, A., Petranka, J.W., and Kats, L.B. 1988. The dynamics of prey refuge use: amodel and tests with sunfish and salamander larvae. Am. Nat. 132(4): 463–483. doi:10.1086/284865.

Skelly, D.K., and Werner, E.E. 1990. Behavioral and life-historical responses oflarval American toads to an odonate predator. Ecology, 71(6): 2313–2322.doi:10.2307/1938642.

Sredl, M.J., and Collins, J.P. 1991. The effect of ontogeny on interspecific interac-tions in larval amphibians. Ecology, 72(6): 2232–2239. doi:10.2307/1941573.

Stenhouse, S.L. 1985. Interdemic variation in predation on salamander larvae.Ecology, 66(6): 1706–1717. doi:10.2307/2937366.

Tarr, T.L., and Babbitt, K.J. 2002. Effects of habitat complexity and predatoridentity on predation of Rana clamitans larvae. Amphib.-Reptilia, 23: 13–20.doi:10.1163/156853802320877591.

Tyler, T., Liss, W.J., Ganio, L.M., Larson, G.L., Hoffman, R., Deimling, E., andLomnicky, G. 1998a. Interaction between introduced trout and larval sala-manders (Ambystoma macrodactylum) in high-elevation lakes. Conserv. Biol.12(1): 94–105. doi:10.1046/j.1523-1739.1998.96274.x.

Tyler, T.J., Liss, W.J., Hoffman, R.L., and Ganio, L.M. 1998b. Experimental analysisof trout effects on survival, growth, and habitat use of two species of amby-stomatid salamanders. J. Herpetol. 32(3): 345–349. doi:10.2307/1565448.

Van Buskirk, J., and Saxer, G. 2001. Delayed costs of an induced defense intadpoles? Morphology, hopping, and development rate at metamorphosis.Evolution, 55(4): 821–829. doi:10.1111/j.0014-3812.2001.tb00817.x.

Welsh, H.H., Pope, K.L., and Boiano, D. 2006. Sub-alpine amphibian distributionsrelated to species palatability to non-native salmonids in the Klamath moun-tains of northern California. Divers. Distrib. 12(3): 298–309. doi:10.1111/j.1366-9516.2006.00254.x.

Werner, E.E., Gilliam, J.F., Hall, D.J., and Mittelbach, G.G. 1983. An experimentaltest of the effects of predation risk on habitat use in fish. Ecology, 64(6):1540–1548. doi:10.2307/1937508.

Wilbur, H.M., and Collins, J.P. 1973. Ecological aspects of amphibian metamor-phosis: nonnormal distributions of competitive ability reflect selection forfacultative metamorphosis. Science, 182(4119): 1305–1314. doi:10.1126/science.182.4119.1305. PMID:17733097.

Wildy, E.L., Chivers, D.P., and Blaustein, A.R. 1999. Shifts in life-history traits asa response to cannibalism in larval long-toed salamanders (Ambystomamacrodactylum). J. Chem. Ecol. 25(10): 2337–2346. doi:10.1023/A:1020830125611.

Zuur, A., Ieno, E.N., Walker, N., Saveliev, A.A., and Smith, G.M. 2009. Mixedeffects models and extensions in ecology with R. Springer-Verlag, New York.

Kenison et al. 249

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Purd

ue U

niv.

Lib

. TSS

on

03/2

3/16

For

pers

onal

use

onl

y.